PDF to JSON API

Turn a PDF into a structured Document Model — element types, text, page numbers and per-element bounding boxes — with one call.

When you need to program against a document rather than just read it, you want structured JSON: every element typed, positioned, and traceable. PDFjet's /extract/json returns a stable Document Model — headings, paragraphs, lists and tables, each with text, a 1-based page number, and a bounding box — so you can build citations, highlight source regions, or drive downstream automation.

It's deterministic and runs on-server, one call, no SDK. Tables include a structure-confidence score so you know which ones to trust, and you can drop the bounding boxes with ?provenance=false when you want a leaner payload.

curl -X POST https://pdfjet.dev/extract/json \
  -H "Authorization: Bearer pj_live_..." \
  -F [email protected]

A stable, documented Document Model

Each page returns its dimensions and an ordered list of blocks. Every block has a type (heading, paragraph, list, table), its text or items, a 1-based page number, and a bounding box in page points. Tables include rows, per-cell coordinates, and a 0-1 confidence score that flags grids worth a human glance.

Provenance you can switch off

Bounding boxes power citations and source-highlighting, but you don't always need them. Add ?provenance=false for a leaner text-plus-structure payload when you're just after the content.

Want specific fields instead? Use schema-guided extraction

If you know exactly which fields you want — invoice number, total, due date — send a JSON schema to /extract/fields and get each value back typed, with page and bounding-box provenance and a confidence score. Deterministic and fully on-server, no third-party model.

curl -X POST https://pdfjet.dev/extract/fields \
  -H "Authorization: Bearer pj_live_..." \
  -F [email protected] \
  -F 'schema={"invoice_number":"string","total":"number","due_date":"date"}'

FAQ